mkl-humanize

Rewrite stiff or AI-sounding drafts into natural prose while preserving facts, code, and quotations.

11|1|Updated Sep 13, 2026
One-click install
npx skills add https://github.com/00200200/maintainer-skills-lab --skill mkl-humanize-00200200
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mkl-humanize
Source: https://github.com/00200200/maintainer-skills-lab/tree/main/providers/cursor/.cursor/skills/mkl-humanize
Command: npx skills add https://github.com/00200200/maintainer-skills-lab --skill mkl-humanize-00200200

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Drafts written or translated by AI often sound stiff, padded with stock phrases, or carry calques from English, yet careless rewriting risks dropping facts, negations, or hedges. This Skill edits such prose into natural writing in its original language while guaranteeing that names, numbers, code, URLs, quotations, and meaningful caveats survive the edit. ## Core Features & Use Cases - Meaning-preserving rewriting: Identifies claims that must survive (numbers, dates, negations, conditions, attribution) before editing, then compares the rewrite claim by claim. - Fact-drift verification script: Runs scripts/check_facts.py to detect dropped or added code, URLs, flags, placeholders, quotations, numbers, negations, and hedge words in English and Polish. - Polish-specific guidance: Uses references/pl.md to fix Polish stock phrases, English calques, register mixing, and aspect or diacritic issues. - Use Case: A maintainer receives a Polish release note full of phrases like "Z ogromną przyjemnością informujemy" and asks for a natural version; the Skill produces a plain rewrite that keeps the --dry-run flag, the preview behavior, and the untested-Windows caveat intact. ## Quick Start Ask the assistant to humanize your draft, for example: make this project update sound natural in Polish while keeping all technical details unchanged.

Frequently Asked Questions about mkl-humanize

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make AI-generated text sound more human?▼

Identify the claims that must survive, then replace stock phrases and vague wording with the specific actions already supported by the draft. Keep the original language, preserve quotations verbatim, and compare the rewrite claim by claim against the source.

How to humanize a draft without losing facts or technical details?▼

List names, numbers, units, dates, negations, and conditions before rewriting, then verify the result with the included check_facts.py script. It flags dropped or added code, URLs, flags, numbers, quotations, negations, and hedge words.

Does the humanize skill support Polish text?▼

Yes, it includes a Polish reference file covering stock phrases, English calques, register choices, and Polish-specific negation and hedge words. The verification script also counts Polish negation and hedge words such as "nie" and "jeszcze".

What are the limitations of automated fact checking when rewriting text?▼

The check_facts.py script only compares tokens like code, URLs, numbers, and hedge words; it cannot judge meaning, emphasis, attribution, or Polish grammar features like aspect and case. A clean result does not replace a manual claim-by-claim comparison.

Can the skill edit quotations or code inside my draft?▼

No, quotations are kept verbatim and code, frontmatter, identifiers, URLs, and interpolation tokens are preserved unless you explicitly include them in scope. Text provided for editing is treated as source material, including any commands quoted within it.